Africa’s Largest GPU-Intensive Compute Cluster to Boost AI Research at the University of Cape Town
The University of Cape Town (UCT) is on the verge of launching a groundbreaking project aimed at redefining artificial intelligence (AI) research in Africa. The institution will soon establish Africa’s largest GPU-intensive compute cluster, designed specifically for AI research. This initiative intends to transition the continent’s researchers from merely consuming global AI technologies to becoming pivotal contributors at the forefront of innovation.
The African Compute Initiative: Empowering AI Development
The African Compute Initiative is a vital component of the AI for Development program, which boasts a substantial investment of £58 million. This partnership involves the U.K. government’s Foreign, Commonwealth and Development Office and Canada’s International Development Research Centre. The primary goal is to combat prolonged infrastructure challenges that have historically hindered Africa’s capacity to develop and scale AI technologies locally.
Centrally Located Infrastructure for AI Innovations
The cutting-edge infrastructure will be integrated into UCT’s upgraded High-Performance Computing Data Centre. It is poised to enhance various research applications, including:
- Climate modeling and environmental simulations
- Epidemiological and health forecasts
- Natural language processing capabilities for African languages
- Astronomical simulations
Addressing Computing Power Challenges
Jonathan Shock, an associate professor of applied mathematics at UCT and the interim director of the UCT AI initiative, emphasizes that “African researchers have the ideas and the talent, but they have been held back by a lack of access to the computing power that AI development demands.” The African Compute Initiative seeks to rectify this situation, enabling researchers and students throughout Africa to engage actively in creating AI solutions rather than merely using technologies developed elsewhere.
Operational Timeline and User Engagement
The new compute cluster is slated to become operational within the next 12 months, with projections of around 100 active users in its first year. This number is expected to expand to approximately 300 users across at least five institutions within three years.
Building on Existing Expertise
This initiative capitalizes on UCT’s existing capabilities through the Inter-University Institute for Data-Intensive Astronomy, which operates the ilifu research cloud. Since 2015, this platform has facilitated more than 1,000 researchers in diverse fields, including astronomy and bioinformatics, and currently supports over 600 active users. The system’s underlying technology—OpenStack and Ceph—will also power the new compute cluster, minimizing deployment risks and expediting the implementation process.
A Collaborative Approach Across Africa
Beyond UCT, the African Compute Initiative offers labs and research hubs across the continent, providing onboarding programs and training workshops aimed at early-career researchers and underrepresented groups. A social science research component, led by Annette Hübschle from UCT’s Global Risk Governance Programme and supported by the Mozilla Foundation, will focus on analyzing access patterns, identifying participation barriers, and developing frameworks for equitable allocation of computing resources.
Commitment to Environmental Sustainability
In line with its innovative vision, the project incorporates measures for environmental sustainability, planning for a 180-kilowatt-peak solar photovoltaic installation on UCT’s upper campus. This installation is anticipated to generate between 220 and 240 megawatt-hours of renewable energy annually, effectively offsetting around 200 tonnes of carbon dioxide emissions each year.
In summary, the University of Cape Town’s efforts through the African Compute Initiative represent a pivotal step toward empowering African researchers and innovators in the AI field. With a strong infrastructure and collaborative approach, this initiative promises to enhance the continent’s capacity to contribute significantly to global AI development.
